3336 - Spatial Immune Phenotyping Using Histo-Immuno-Radiomics in Patients with Biochemical Recurrence Following Radical Prostatectomy
Presenter(s)
O. M. Oderinde1,2, R. Chimmula3, V. Nair4, J. Holmes5, and C. Bahler6; 1Advanced Molecular Imaging in Radiotherapy (AdMIRe) Research Laboratory, School of Health Sciences, Purdue University, West Lafayette, IN, 2Department of Radiation Oncology, Indiana University School of Medicine, Indianapolis, IN, 3Advanced Molecular Imaging in Radiotherapy (AdMIRe) Research Laboratory, School of Health Sciences, Purdue University, Indianapolis, IN, 4Department of Pathology, Indiana University School of Medicine, Indianapolis, IN, Indianapolis, IN, 5Department of Radiation Oncology, Indiana University School of Medicine, Indianapolis,IN, Indianapolis, IN, 6Department of Urology, Indiana University School of Medicine, Indianapolis, IN
Purpose/Objective(s): Biochemical recurrence (BCR) following radical prostatectomy (RP) remains a major clinical challenge, with uncertainty regarding optimal timing and patient selection for salvage radiotherapy (SRT). While prostate cancer (PCa) is often considered immunologically “cold,” the spatial immune microenvironment in PCa is poorly characterized. We hypothesized that histo-immuno-radiomic features derived from CD8 and CD163 immunostained whole-slide images could identify immune phenotypes associated with early recurrence, potentially informing SRT stratification.
Materials/Methods: Four tumor-containing Hematoxylin and Eosin (H&E) RP blocks were selected per patient from 14 individuals who developed biochemical recurrence (BCR) following RP. Serial sections were immunostained for lymphocytes (CD8) and macrophages (CD163) and digitally scanned at high resolution. Tumor-enriched tiles were extracted from each slide, and quantitative radiomic features were computed at the tile level. Features were then aggregated to patient-level summaries using central tendency and heterogeneity metrics. From these data, four biologically informed composite indices were constructed: a CD8 Activation Score (reflecting low entropy, high energy, and high coarseness), a CD163 Suppression Score (reflecting high entropy, high zone variance, and high busyness), an Immune Suppression Composite (CD163 minus CD8 activation), and an H&E Morphologic Chaos Index. Early recurrence was defined as biochemical recurrence occurring within 24 months of surgery, and differences between early and later recurrence groups were evaluated using Mann–Whitney U testing.
Results: Five patients experienced early recurrence (=24 months) and nine had later recurrence. Early recurrence tumors demonstrated a numerically higher CD8 Activation Score (mean 0.52 vs -0.29, p=0.18), suggesting increased spatially organized CD8 architecture, although this did not reach statistical significance. CD163 Suppression Scores were similar between groups. The Immune Suppression Composite and H&E Chaos Index showed no significant separation. These findings suggest that early recurrence may not be driven by macrophage-dominant suppression but rather by a distinct CD8 spatial phenotype.
Conclusion: In this pilot cohort, spatial CD8 radiomic activation demonstrated directional association with early biochemical recurrence following RP. These findings suggest that early-recurrent tumors may harbor an organized yet potentially ineffective CD8 microenvironment, raising the possibility that immune-informed stratification could guide timing and intensification of salvage radiotherapy. Prospective validation in larger cohorts is warranted.